LT5903 Chap.2 Speech Production and Perception
Speech Production and Perception
Three views of one speech event
Articulatory phonetics describes the gesture, acoustic phonetics describes the signal and auditory phonetics describes perception. The chapter uses IPA to replace unreliable spelling, classifies consonants by voicing, place and manner, and classifies vowels by tongue height, advancement, rounding and tenseness.
It then reads time, frequency and amplitude from spectrograms, connects formants to the source-filter model and compares general-auditory with motor accounts of speech perception.
Keep spelling separate from sound
Orthography may use one letter for several sounds or several letters for one sound. Transcribe the token actually described rather than converting the written form mechanically.
First identify whether airflow is obstructed enough to form a consonant or remains relatively open as a vowel. Then assign only features supported by the prompt. A visible closure can justify a stop manner, vibration can justify voicing and the active and passive articulators determine place.
If a cue is absent, leave the feature open instead of guessing from an English spelling.
Classify consonants as a feature bundle
Voicing describes vocal-fold vibration. Place describes where the strongest constriction occurs, and manner describes how airflow is shaped. These dimensions combine, so changing one need not change the others.
Compare pairs systematically: hold place and manner constant to isolate voicing, or hold voicing and manner constant to isolate place. Nasals require an oral closure with a lowered velum; fricatives require a narrow channel that sustains turbulence; affricates combine a stop-like closure with a fricative release.
State the evidence for every feature rather than giving only an IPA symbol.
Read vowels and formants relationally
Vowels are organised by tongue height, frontness or backness, lip rounding and sometimes tenseness. The acoustic signal reflects filtering by the vocal tract. The first formant generally varies inversely with vowel height, while the second formant relates strongly to advancement.
Consequently, a higher F1 supports a lower vowel and a higher F2 supports a fronter vowel, all else equal. Treat those relations as measurements, not as the formants themselves being tongue positions.
Speaker anatomy and context can shift absolute values, so comparisons within an appropriate set are safer.
Locate evidence on a spectrogram
The horizontal dimension represents time, the vertical dimension frequency and darkness or intensity represents energy. A stop closure may appear as reduced energy followed by a burst, while a fricative often shows sustained aperiodic energy.
Regular vertical striations can indicate periodic voicing. Formant bands provide resonant information for vowels. Read the axes before interpreting a patch, and connect a visual cue to an articulatory event.
A diagram does not automatically decide phonological status; it describes the physical token that a later distributional analysis may classify.
Compare perception theories through predictions
A general-auditory account emphasizes perceptual mechanisms also used outside speech, whereas a motor account emphasizes recovery of articulatory gestures. Do not reduce the contrast to ears versus mouth.
Ask what each account predicts when acoustic cues vary but perceived gestures remain stable, or when non-speech stimuli share relevant auditory structure. Evidence can favour one explanation for a task without settling every aspect of speech perception. Finish by naming the stimulus manipulation, measured response and theoretical inference separately.
What this chapter covers
- 01
Articulatory, acoustic and auditory phonetics
- 02
IPA consonant descriptions
- 03
Vowel-space features
- 04
Spectrogram axes and formants
- 05
Competing accounts of speech perception
Worked example · free
Build a consonant description from cues
- 2Infer voicing.
- 2Infer place.
- 2Infer manner.
Key terms
- Voicing
- Vibration of the vocal folds during a sound.
- Place Of Articulation
- The location of the main vocal-tract constriction.
- Manner Of Articulation
- The way airflow is constricted.
- Formant
- A frequency region amplified by vocal-tract filtering.
- Spectrogram
- A display of frequency and amplitude over time.
- Source-Filter Model
- A model separating vocal-fold excitation from vocal-tract resonance.
Speech Production and Perception FAQ
Why is ordinary spelling unreliable for phonetic analysis?
One spelling can represent different sounds and one sound can be represented by different spellings. IPA supplies a more stable sound-symbol relation, letting an answer describe what was produced rather than what letters happen to appear in the word.
What should I read first on a spectrogram?
Orient the axes before interpreting any pattern: time is horizontal, frequency is vertical and darkness represents amplitude. Then locate sustained formants, closures, release bursts or frication. Naming the visible cue prevents a label from becoming guesswork.
How do motor and auditory accounts differ?
A motor account treats intended articulatory gestures as the objects recovered in perception. A general-auditory account treats acoustic events as the objects and relies on broadly available perceptual learning. Evidence should be evaluated by which mechanism it distinguishes.
Exam move
Practise descriptions in a fixed order: transcription, voicing, place and manner for consonants; height, advancement, rounding and tenseness for vowels. For acoustic problems, label all axes before interpreting a trace. Pair each gesture with an expected acoustic cue, then reverse the exercise by inferring a cautious articulation from a cue.
Keep a column for alternative explanations because no single dark band or burst should carry more interpretive weight than the measurement allows.
Working through Speech Production and Perception in LT5903? Sia is AskSia’s AI Arts and Humanities tutor — ask any LT5903 Speech Production and Perception question and get a clear, step-by-step explanation grounded in how LT5903 is taught and assessed. Read this chapter free, then take your hardest questions to Sia.
